fix(engine): play-from-exile permission is consultable from the graveyard for milled cards (#751)

[matthewevans]

Issue

Fixes #751. Ark of Hunger (and the class of "mill a card, then you may play it" cards — CR 701.17d) grants a PlayFromExile permission on a card that lands in the graveyard, but the engine's "can this object enter the cast path?" consult sites were gated to Zone::Exile only. The milled card carried a live permission yet was never castable/playable.

Fix

Introduces a separate predicate play_from_exile_object_in_cast_path(obj) = matches!(obj.zone, Zone::Exile | Zone::Graveyard) && !is_land, used at the three object-tagged PlayFromExile consult sites plus a new graveyard cast-surface branch and a new graveyard land sweep.

The original exile_object_can_enter_cast_path stays Zone::Exile-only, and the two static-ExileCastPermission callers are byte-for-byte unchanged — a card milled into a graveyard must not reach the static path (exile-only by CR 113.6b). Lands are excluded from the cast path (CR 305.1) and routed through the land-play sweep.

Tests

crates/engine/tests/integration/ark_of_hunger_play_from_graveyard_751.rs drives the real Mill→grant→cast pipeline:

• milled_card_is_castable_from_graveyard (discriminating — flips false→true on the fix)
• milled_land_is_playable_via_land_path_not_cast_path (CR 305.1 carve-out)
• exiled_card_with_play_permission_stays_on_exile_path (Advanced Reconstruction regression — gate is exile-OR-graveyard, never any-zone)
• graveyard_play_permission_is_pruned_at_end_of_turn (CR 514.2)

All four PASS in local Tilt test-engine. Independent /review-impl: APPROVED, zero findings. CR annotations (113.6b, 305.1, 514.2, 601.2a, 604.2, 701.17a, 701.17d, 715.3d) grep-verified against the Comprehensive Rules.

[gemini-code-assist]

Code Review

This pull request introduces a regression test for GitHub issue #751 (Ark of Hunger / Tablet of Discovery) to verify that milled cards carrying play permissions can be correctly played or cast from the graveyard and that these permissions are properly pruned at the end of the turn. The feedback suggests refactoring the test helper mill_one_and_grant to replace the is_land: bool parameter with a typed enum, adhering to Style Guide Rule R2 to avoid boolean fields and improve type safety.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

The is_land: bool parameter should be replaced with a typed enum to better express the card's properties.

Why it matters: Using a bool is contrary to the spirit of Style Guide R2 ('No bool fields'), which favors typed enums for expressiveness and safety. It also leads to passing a meaningless ManaCost::NoCost for lands.

Suggested fix:

1. Introduce a new enum to represent the milled card's properties:

enum MilledCardInfo {
    Land,
    Spell(ManaCost),
}

2. Update the mill_one_and_grant function to accept this enum. The function body (lines 83-88) should be changed to a match statement on this new enum.
3. Update the call sites accordingly, e.g., MilledCardInfo::Land or MilledCardInfo::Spell(cost).

fn mill_one_and_grant(
    runner: &mut GameRunner,
    library_card_name: &str,
    card_info: MilledCardInfo,
) -> engine::types::identifiers::ObjectId {

References

1. Rule R2 states that bool fields should be avoided in favor of typed enums to better express the design space. While this is a function parameter and not a field, the principle of using typed enums for clarity and safety applies. ^(link)